Design of a Semantic Video Object Analysis and Retrieval System and Its Application to Golf Sports Analysis

瀏覽統計 Email 通知 RSS Feed

簡歷

基本資料

Project title

Code/計畫編號

NSC93-2213-E327-002

Translated Name/計畫中文名

語意式物件為主的視訊資料檢索及分析系統之設計---以高爾夫球運動分析為例

Project Coordinator/計畫主持人

Shyi-Chyi Cheng

Funding Organization/主管機關

National Science and Technology Council

Department/Unit

Department of Computer and Communication Engineering, NKUST

Website

https://www.grb.gov.tw/search/planDetail?id=1024983

Year

2004

Start date/計畫起

01-08-2004

Expected Completion/計畫迄

31-07-2005

Bugetid/研究經費

719千元

ResearchField/研究領域

資訊科學--軟體

本計畫提出一個新的模型為主之語意式視訊物件切割及追蹤方法，並據以應用到視訊內容導向的相關應用上。首先，我們提取在視訊序列中兩個連續畫面之間的移動區域組成的前景物件,以便標示視訊序列的畫面內容。為了這目的,基於矩量保持邊緣偵測方法，我們結合包括顏色、像素運動等若干種類的資訊，提出一結合時間、空間的新型區域分割方法，包含在前景物件內的區域，又稱為前景區塊，一定是一個運動中的區域。接著，本計畫提出一結合一般化Hough轉換之模型物件的物件追蹤演算法。以高爾球為例，本計畫分析所得到領域知識從視訊片段內各畫面，擷取一般特徵向量及運動特徵向量，透過追蹤視訊序列中兩個連續畫面之間的前景內的語意物件的特徵變化，我們提出一以動態規劃為基礎的語意式視訊片段相似度計算方法，並據以建構一語意式視訊檢索系統。物件為基礎的視訊資料表示方法, 有利於建構內容導向系統統的相關功能及提高編碼效率, 物件的可用特徵包括形狀、顏色、紋路和移動向量等資訊，但最要的是必須包含語意特徵。植基於這個視訊分割平台,我們提出一符合語意的視訊物件檢索系統。初步的實驗結果顯示本計畫所提出的視訊分割方法及物件追蹤的效率效果顯著。 This project presents a new semantic video object segmentation method and an object tracking scheme for content-based applications. The foreground object, consists of a number of moving regions between two successive frames in a video sequence is first extracted in order to identify the content of the video segment. For this purpose, we propose a novel spatial-temporal segmentation method as a general segmentation algorithm combining several types of information including color and motion using a moment-preserving edge detection algorithm. A region within a foreground object is called as a foreground region, which is characterized as a moving uniform region. A foreground model is then used to cluster the foreground regions into a set of meaningful video objects by the generalized Hough transform. An algorithm for object tracking based also on the generalized Hough transform is also included in order to recognize feature variations of video objects. As an example, the proposed method is applied to analyze golf sports motions of the users of the system. Moreover, a semantic based similarity measurement of two video segment is proposed by dynamic programming. Object-based video data representations enable content-based functionality, as well as high-coding efficiency, by taking into account shape, color, motion information of moving objects, and especially the semantic features. Building upon the segmentation framework, we then present a unique object-based query system for semantic video object. Primary experimental results are presented to demonstrate the performance of the new method in terms of better segmentation and computational efficiency.

Keyword(s)

彩色視訊影像
語意物件切割
一般化Hough 轉換
物件追蹤
物件式視訊檢索
高爾球運動分析
Semantic Video Object Segmentation
Moment-preserving Edge Detection
Object tracking
Golf Sports Analysis
Object-based video retrieval

DSpace CRIS

Design of a Semantic Video Object Analysis and Retrieval System and Its Application to Golf Sports Analysis

基本資料

Description